A non-vented non-steady state flow-through chamber and a non-vented non-steady state non-flow-through chamber technique were used to measure CO2 efflux of a young Scots pine forest on a fertile till soil in southern Finland. Soil temperature, soil moisture and soil CO2 concentration were measured concurrently with CO2 efflux for two and a half successive years. The CO2 efflux showed a seasonal pattern, effluxes ranging from low 0.0–0.1 g CO2 m−2 h−1 in winter to peak values of 2.3 g CO2 m−2 h−1 occurring in late June and in July. The daily average effluxes in July measured by flow through chambers were 1.23 and 0.98 g CO2 m−2 h−1 in 1998 and 1999, respectively. The annual accumulated CO2 efflux was 3117 and 3326 g CO2 m−2 in 1998 and 1999, respectively. The spatial variation in CO2 efflux was high (CV 0.18–0.45) and increased with increasing efflux. Soil air CO2 concentration showed similar seasonal pattern the peak concentrations occurring in July–August. The CO2 concentrations ranged from 580 to 780 µmol mol−1 in the humus layer to 13 620–14 470 µmol mol−1 in the C-horizon. In winter the soil air CO2 concentrations were lower, especially in deeper soil layers. Drought decreased CO2 efflux and soil air CO2 concentration. The in situ comparison on forest soil between the chamber methods showed the non-flow-through chamber to give ∼∼50% lower efflux values than that of the flow-through chamber. When calibrated against known CO2 efflux ranging from 0.4 to 0.8 g CO2 m−2 h−1 generated with a diffusion box method developed by Widén and Lindroth [Acta Universitatis Agriculturae Suecia Silvestria, 2001], the flow-through chamber gave equal effluxes at the lower end of the calibration range, but overestimated high effluxes by 20%. Non-flow-through chamber underestimated the CO2 efflux by 30%.